Structural equation models (SEM) belong to latent model analysis. This type of analysis is used to deal with several difficult modelling challenges, including cases in which some variables of interest are unobservable or latent and are measured using one or more exogenous variables. The objective of the present research is to investigate whether the case of the unobserved driving performance can be assessed through this type of analysis. For this purpose, a large driving simulator experiment was carried out, in which 95 participants from all age groups were asked to drive under different types of distraction (no distraction, conversation with passenger, mobile phone use) in rural/urban road environment, in low/high traffic. Data collected from the driving simulator experiment include 16 continuous driving performance parameters such as longitudinal control measures (mean speed, headways, etc.), lateral control measures (lateral position, standard deviation of lateral position etc.), the reaction time of the driver at unexpected incidents and other driving performance parameters. Then, in the framework of the statistical analyses, latent analysis is implemented, including 2 structural equation models, in which the latent variable reflects the unobserved driving performance of the participants, and is based on several driving performance parameters. Moreover, in the structural part of the model the effect of several variables includes distraction sources, area type (urban/rural area), traffic conditions (low/high traffic) and driver characteristics (age, gender, driving experience) on driving performance is estimated. Results indicate that the selection of the specific measures that define overall performance should be guided by a rule of representativeness between the selected variables, as in the present structural equation models the unobserved driving performance are defined based on a longitudinal measure, a lateral measure and a time related measure.
|Tags||driver behaviour, driving simulator, machine learning, statistical modelling|